Unhelpful Statistics

29 Oct 2019 at 20:57 // under misc

The headline of a BBC news article I spotted at lunchtime read: Homicide numbers in Scotland show slight rise

The sub heading read:

A total of 61 people were killed across the country last year - up 2% on the previous 12 months.

Why on earth would you bother putting a percentage in those numbers. Why not say that last year there was 59¹. When you say "up 2%" and "slight rise", you would expect that there is a small but real change in what is happening. But a change in 2 deaths, that is just noise, random chance.

If there were 600 deaths and this had increased to 612, then I think you can summarise the numbers by talking about a 2% rise. But if the number of people involved is less than 100, I really don't think converting it to a percentage is useful, and just makes things more complicated.

The article does at least compare this year's figure with historical numbers (it was the third lowest figure in the last 40 years). It may have been beneficial to also compare the murder rate with other death rates such as from traffic accidents (160 deaths in 2018²) or to compare with other countries. Of course to get these other figures, someone would actually have to research publications by a different department.

I was initially annoyed at the BBC for producing these unhelpful statistics though I see that the source publication by the scottish government also included these numbers. The government's report at least mentions that it is an increase of one case and two victims right beside the percentage increases.

In any case, both organisations should know better.

1. It is actually up 3%. Someone at the BBC got the percentage increase for ‘number of cases’ and ‘number of deaths’ mixed up. ↩
2. Source: Transport Scotland ↩

Go Top